This title appears in the Scientific Report :
1999
Please use the identifier:
http://hdl.handle.net/2128/20369 in citations.
Parallele numerische Algorithmen und mathematische Theorie zur Magnetenzephalographie
Parallele numerische Algorithmen und mathematische Theorie zur Magnetenzephalographie
Magnetoencephalography is a method for functional brain imaging It is possible to estimate the location of sources of biomagnetic activity in the brain using di erent reconstruction methods Unfortunately the reconstruction methods usually require a large amount of computation which takes time It is...
Saved in:
Personal Name(s): | Beucker, R. |
---|---|
Contributing Institute: |
Zentralinstitut für Angewandte Mathematik; ZAM |
Imprint: |
Jülich
Forschungszentrum, Zentralbibliothek
1999
|
Dissertation Note: |
Aachen, Techn. Hochsch., Diss., 1999 |
Document Type: |
Book Dissertation / PhD Thesis |
Research Program: |
Mathematische Methoden und Algorithmen für neue Rechnerstrukturen |
Series Title: |
Berichte des Forschungszentrums Jülich
3630 |
Subject (ZB): | |
Link: |
OpenAccess OpenAccess |
Publikationsportal JuSER |
Magnetoencephalography is a method for functional brain imaging It is possible to estimate the location of sources of biomagnetic activity in the brain using di erent reconstruction methods Unfortunately the reconstruction methods usually require a large amount of computation which takes time It is important to get information about the quality of the data directly after the measurement in order to know if the data is worth processing further To get an idea of the data quality in a timely manner a parallel sliding window approach based on a multiple dipole reconstruction method is developed for the massively parallel computer system CRAY T E of the research centre Julich Furthermore a special kind of minimumnorm method is proposed which has some advantages with respect to stability against noise The algorithm for the calculation of the solution is implemented on the CRAY J of the research centre Julich The problem of determing the number of active sources for the sliding window approach is done with a statistical test and the necessary improvement of the noise conditions for the application of the test is solved by a multichannel prewhitening method |